Abstract: To analyze and evaluate the environmental self-purification function of river water, the chemical and hydro-meteorological data have been collected over the past 6 years in the Shigenobu river of Shikoku island, Japan. Generally, the problems of an adequate understanding and description of complex processes of pollutant circulation and water quality dynamics in rivers can be effectively solved by applying the water quality mathematical models. This approach allows taking into account, analyzing and ranking numerous interacting hydrological, meteorological and biological factors, impact of natural and anthropogenic sources of pollution. But in some cases the hydrological, meteorological and biological information are unavailable or incomplete. Thus this study presents the results of a study that examined the application of Artificial Neural Network (ANN) model to simulate the river chemical mass transport using the indicator of dissolved oxygen. Artificial Neural Network model is proved to behave well with the measured data. This study demonstrated that it was feasible to assemble and deploy ANN model to predict the dissolved oxygen under the condition of incomplete information.